IS TRUTH EFFABLE?
Pascal Engel
Université Paris IV
Preliminary version of “Is Truth effable?”
L. Hahn, ed. The philosophy of Jaakko Hintikka
Library of Living philosophers, La salle, Open court, 2005
1.
Professor Hintikka’s work has so much breadth and scope that it is tempting to
think of him as a contemporary Leibniz, if he had not warned us against the nostalgia
for systematic philosophy:
“One upon a time a serious philosopher was expected to produce a system of
his own. Twentieth century philosophers are likely to find this expectation not only
old-fashioned but more than a little ridiculous. Professional standards of clarity and
argumentation are in our time and age such that it is hard enough to produce first
class work in one limited area of philosophy. Hence the idea of marshalling deep
philosophical thought on command about each and every philosophical topic seems
to be about as relevant to us as the principles of Napoleonic warfare are to global
strategy (Hintikka 1987: 9)
Nevertheless, Hintikka’s work is very systematic, both in his constant defense and
illustration of the model-theoretic tradition in logic, in its application to so many fields
as modal logic, epistemic logic, logical semantics, and in his analysis of the “ultimate
presuppositions” (a term he borrows from Collingwood) which lie behind the
traditions of thought that he examines. One such great presupposition, which he has
brought into light better than anyone else, is the universalist assumption in logic and
the thesis of the ineffability of semantics. Hintikka argues that his own gametheoretical approach in logic and in semantics allow us to claim that semantics is not
ineffable, and to reject the universality assumption. In an illuminating series of essays
2
1
, he has analysed how the thesis of the ineffability of semantics affects our
understanding of the concept of truth, which becomes itself ineffable. Now the
question we may ask is this: what conception of truth emerges from Hintikka’s
rejection of the ineffability thesis? I want to try to characterise the philosophical
concept of truth which underlies Hintikka’s analyses, and in particular to confront it
with the “deflationary” and “minimalist” conceptions of truth which have recetly
occupied the front scene. I shall try to show that, in spite of his criticism of
Wittgenstein’s conceptions of truth and semantics, Hintikka is still very close to
Witgenstein’s conception of meaning as use, and that Hintikka’s view can still be
characterised as a form of minimalism about truth and meaning.
2.
Let us try first to characterise the ineffability thesis about truth and semantics,
starting, as Hintikka most often does, from Wittgenstein’s views. In a famous passage
of the Philosophical Remarks, Wittgenstein says :
In a certain sense, the use of language is something that cannot be taught,
i.e I cannot use language to teach it in the way in which language could be used
to teach someone to play the piano. And that of course is just another way of
saying that I cannot use language to get outside language” (Wittgenstein, 1975,
p.54)
There are at least two lines of thought, closely associated to each other, in such
passages. The first is the thesis of the ineffability of semantics proper, that semantic
relations cannot be established from without, but only from within a language, and
actually from within our language, the language that we speak. The ineffability of
semantics goes hand in hand with the thesis of the universality of language, that
language is the universal medium of communication and thought. Actually, as
Hintikka notes, the former assumption is stronger, and entails the second:
1
Especially those in Lingua Universalis vs Calculus ratiocinator: an ultimate presupposition of Twentieth
Century Philosophy, Selected papers, vol 2., Dordrecht, Kluwer 1997, but in many other places as well. Hintikka
has shown that the idea of language as a universal medium plays a structuring role in much of contemporary
philosophy, and in this sense it is comparable to the role played, in the context of philosophical thought about
modality, to the principle of plenitude. I tried once to examine the status of this principle in Engel 1988.
3
“If semantics is ineffable, it makes no sense to try to speak in our language of a
situation in which the expressions of one’s language would have meanings
different from what they in fact have. In other words, if semantics is ineffable, it
makes no sense to try to say or to assume, by using my actual home language,
that there are languages other than it or that I am changing the semantics of my
language. “A language that I do not understand is no language”, as Wittgenstein
puts its. Hence the ineffability of semantics entails the universality of language.”
(Hintikka 1989, p.23)
The second line of thought present in Wittgenstein’s passage is that “the use of
language cannot be taught”, and that for this reason one cannot use language to teach
the meanings of the expressions of language. Wittgenstein is not here saying that
language cannot be taught, for this would be obviously false, but only that “in a
certain sense” it cannot be taught. Why ? What he seems to say is that any rule for the
use of a word must be expressed in language, and that we cannot use language itself to
explain the rules: in some sense the way the signs are used is primary. This means that
signs can only convey meaning if at some point there is a natural uptake on how they
are used, which cannot be taught. Although it is in a sense just another way of
formulating the ineffability thesis, what Witgenstein says can be formulated thus:
there is no way to explain what it is to know the meaning of a word, and to
understand a language, which would be independent from our knowing already how
words are used. Hence there is no language independent account of what it is to know
meaning. In particular (and this is a familiar theme emerging from Wittgenstein’s
“rule-following” considerations), there is no account of meaning which could invoke
our grasping concepts or propositions, conceived either as psychological entities
(“ideas”, “representations”) or as abstract Platonic entities independent from
language. In other words, there is no priority of a structure of thought - or of a
structure of ontological entities independent from language - over language.
Hintitkka has commented widely upon this idea in Wittgenstein’s writings 2. Following
John Skorupski (1997) who himself adapts a phrase from Dummett, let us call this the
thesis of the priority of language (or the priority thesis, for short): it says that any account of
thought and concepts is intrinsically dependent upon an account of language rules and
2
see in particular Hintikka & Hintikka 1986,ch.1
4
language-understanding. Although, as I have just said, the priority thesis seems to
entail the ineffability thesis (there is no way of formulating the semantics of our
language outside our language), they are not equivalent and the latter does not
necessarily entail the former. It is open to a theorist to defend the view that semantics
in ineffable although thoughts and concepts are independent from language. For
Frege, for instance, thoughts are language-independent entities, although he suscribes
to the ineffability thesis and to the universality assumption. 3
3.
Now, what are the consequences of the thesis of the ineffability of semantics
and of the thesis of the priority of language for the nature of the concept of truth? As
Hintikka shows, the first thesis implies the ineffability of truth, as the main semantic
relation. This line of thought is clearly present in Wittgenstein, when he argues,
against a correspondence theory of truth, that such a theory is impossible, because “it
is impossible to describe the fact which corresponds to a sentence without repeating
the sentence”.4 Because the relationship between sentences and facts, propositions
and reality, cannot be spelled out, truth admits of no other “definition” than this one:
“ For what does a proposition’s “being true” means? “p” is true = p (That is the
answer)”
5
As Hintikka points out commenting this passage, the basis of this Wittgensteinian
view is “the impossibility of expressing in language the conditions of agreement
between a meaningful proposition – a thought – and reality”6 And it this sense it does
not amount to a definition, or to a theory of truth. It is simply the denial that any sort of
such definition or theory could be given.
In so far as one could ascribe a “theory” (and not simply an elucidation) of
truth to Wittgenstein in such passages, it would fall within the category so so-called
3
On this point, see e.g. Hintikka 1981
Vermischte Bemerkungen, tr.engl; Culture and Value Blackwell, Oxford, 1980 p. 10, quoted and translated by
Hintikka 1989 p.24
5
Remarks on the Foundations of Mathematics, Oxford, Blackwell, 1978, 3rd ed. Appendix I, sec 5. quoted by
Hintikka, op cit. p.23.
6
“Is truth ineffable?” op cit, p.23 (the quote is from MS 108,p.265).
4
5
“minimalist” or “deflationist” theories of truth. Actually, Witgenstein is often
considered as a representative of the “redundancy” theory of truth, alongside Ramsey
and Ayer.
The redundancy conception of truth should not be confused with the
disquotational conception. According to the former, truth is a (non genuine) property of
propositions, the meaning of which we already know, whereas according to the latter,
truth is a (non genuine) property of sentences. The role of “true” is to express Tsentences of the form “ ‘ p’ is true
iff p”or infinite disjunctions and conjunctions of sentences of the form: x is true iff (
x = ‘s1’ & s1 ) or (x = ‘s2’ & s2 ) or…. .7 What these views have in common, as Paul
Horwich (1990) as pointed out, is the claim that truth is not a genuine property, but
only a quasi property : truth has no essence, substance, or explanatory role, and so
cannot be explained in terms of such properties or relations as correspondence,
coherence, utility, etc.
Against such minimalist views of truth, Hintikka argues that they belie the
ineffability assumption about semantics. Tarski’s hierarchy of metalanguages is itself
but a version of the view that truth cannot be defined and that the ultimate semantical
relations cannot be spelled out. This is because both disquotational and other
“minimalist” conceptions of truth suffer from the same difficulty as Tarski’s
definitions: they cannot be formulated in the language to which they are supposed to
apply.8 They operate at a purely syntactical level. Indeed, for the minimalist the truth
predicate only obeys the discipline of syntax : it allows us to quote and to disquote
sentences, and to embedd them within propositional attitude and other contexts.
9
Against this, Hintikka argues that truth can be defined, both for a formal and for a
natural language, by dropping Tarski’s hierarchy, through an Independence-Friendly
(IF) logic and a game-theoretical semantics, by developing a theory of truth “in some
7
For various versions of the disquotational conception, see David 1994.
“Defining truth, the whole truth and nothing but the truth”, Selected papers, 2,”, op.cit. p.76
9
This very feature has been called “syntacticalism” by some minimalist theorist. See in particular Crispin Wright
1992
8
6
suitable metalanguage which does not have to be thought as being sharply separated
from the object language itself”10. By contrast,
“…Disquotational treatments of truth are …subject to the standard criticism
which has been levelled at Tarski’s treatment of truth. They do not tell anything about
the way sentences are in fact shown to be true or false. Indeed, when nothing is said
of the language games through which truth and falsity are constituted, it is natural to
resort to disquotational ideas.” (Hinitkka 1991, p.76)
Not only disquotional account of truth do little to illuminate truth, but also
they do little to illuminate meaning. Where famously a number of philosophers,
following Davidson, have hoped to built an empirical theory of meaning on the basis
of theories of truth satisfying (to a certain extent) Tarski’s Convention T, Hintikka has
objected that such accounts fail for a number of simple quantificational sentences
(such as “Any corporal can become a general”) and that the principle of semantic
compositionality upon which they rest fails as well. 11
4.
My concern here is not to examine the specific nature of Hintikka’s proposals
in logic and in model-theoretic semantics. The question that I intend to raise is this:
what kind of analysis of the ordinary concept of truth emerges out of Hintikka’s
analyses, and what kind of philosophical conception of truth does it support? Given
his criticism of minimalist theories of truth, one should expect that Hintikka should
propose some kind of substantive theory of truth, where by “substantive” I mean any
conception of truth which would reject the minimalist view and the ineffability thesis,
and which would claim that truth is a genuine property of some sort which could be
defined and spelled out fully. But which one ? One suggestion which come naturally
to mind could be that Hintikka defends a variety of verificationist theory of truth. As
he says in the passage above when he contrasts the game-theoretical approach with
the disquotational approach to truth, the former does, in a way in which the latter
does not, tell us how sentences are shown to be true or false. In a nuthshell , the game
theoretical analysis of the truth conditions of sentence says that a sentence S is true in
10
11
Hintikka, “Defining Truth, the whole truth and nothing but the truth”, 1997, op cit, p. 87.
see in particular Hintikka 19XX
7
a given model M if an only if there exists a winning strategy for the initial verifier in
the game G(S) when played on M. Now this definition, because of the role played in it
by the notion of verification, has often been compared with an anti-realist conception
of meaning in terms of assertability conditions, such as Dummett’s. But although he
admits that his views are very much “in the spirit of the constructivist way of
thinking”, Hintikka has stressed that his own conception of truth is not
constructivistic or anti-realist.12 Dummett’s famous analogy between the notion of
truth and the notion of winning a game is not the good one if truth is supposed to be
identified in some way with the strategy of verification. This is not so: “The interesting
analogy is between the notion of truth and the existence [my italics] of a winning
strategy.”13 In other terms, the winning strategy is already there, it is not constructed
or created in any sense by the steps in which it consists. So it it not part of the gametheoretical conception that truth conditions should be understood in some sense as
investigation-dependent or as known. In spite of the analogy between the notion of
game and the notion of verification, truth, Hintikka insists, is not known truth; so his
concept of truth is not to be identified with the constructivist’s or to the intuitionist’s
one. Actually Hintikka’s substantive conception of truth seems to be closer to the
realist conception of truth, which presupposes that there are verification transcendent
truth conditions for sentences. This point emerges in particular in Hintikka’s analysis
of such first order sentences with dependent quantifiers as
(1) (∀x) (∃y) F [x, y ]
These are true not only when it is possible to find a “witness” individual y depending
on x such that F [x, y]. But one can find such an individual only if there is a certain
function f such that
(2) (∀x) F [x, f (x)]
12
13
see , The Principles of Mathematics Revisited, p. 210; “Defining truth…, op cit, p.65
Hintikka 1996, p. 27.
8
where f is a Skolem function, namely a choice function of individuals.14 So the truth
condition of (1) is the existence of such a Skolem function. In other terms,
appropriate witness individuals exist for a quantificational sentence S only if there is
an “array” of Skolem functions. The quantifier “there exists” should be understood
here objectually. As Hintikka suggests in a recent essay, one could even use the notion
of a truth-maker, which has been invoked in the context of recent realistic
conceptions of truth 15. When it comes to the establishment of the truth-conditions of
our sentences, Hintikka’s account is genuinely semantical, in the sense in which David
Lewis said famously that “semantics without truth conditions is no semantics”. But
when it comes to seeking truth, Hintikka’s account is constructivistic: it involves the
activities of finding truths. But the first cannot be reduced to the second, and it is only
when one confuses semantical games with interrogative games that one is lead to the
idea that truth is in some way dependent upon our human activities.16
Now, is Hintikka a straighforward realist about truth in the sense of a
correspondence theory? There are several reasons to doubt this. In the first place,
Hintikka has little sympathy , to say the least, for an ontology of facts and states of
affairs, and even less for an ontology of real possible worlds à la Lewis, which are the
entities usually invoked by genuine correspondence theories of truth. In the second
place, he has often reiterated Wittgenstein’s criticism againt such theories.
Commenting upon the difference between his game-theoretical approach and
pragmatist conceptions of truth he says :
“The possibility of a game-theoretical concept of truth which accords with our
natural concept of truth, together with the distinction between semantical ( truthconditioning) and interrogative ‘truth-seeking) games also has profound philosophical
repercussions. For one thing, it shows what is true and what is false in pragmatist
conceptions of truth. What is true is that to speak of truth is not to speak of an
independently existing correspondence relations between language and the world.
There are no such relations. Or, as Wittgenstein puts it, the correspondence between
14
see Hintikka 2001.
Hintikka 2002, manuscript. I thank professor Hintikka for having allowed me to read this manuscript. As
Hintikka notes, however, this cannot really be considered as equivalent to, say, David Armstrong’s notion of
truth-making ( see e.g. Armstrong , 1997), for according to Armstrong sentences are made true by a host of
different kinds of “truthmakers”: objects, properties, relations, states of affairs and facts .
16
see e.g. The Principles of Mathematics Revisited, p. 42-44
15
9
language and the world can be established only by the use of our language – that is by
semantical games. Truth is literally constituted by certain human rule-governed
activities. What is false in pragmatist ideas about truth is the claim that the relevant
activities are the activities by which we typically find out what is true – that is to say
verify, falsify, confirm, disconfirm and so forth, our propositions. This claim is based
on overlooking the all-important distinction between truth establishing games (that is
semantical games) and truth seeking games (that is interrogative, or perhaps other
epistemic games)….Our actual truth seeking practices, whether or not they are
relative to historical era, epistemic or scientific community, social class of gender, are
not constitutive of our concept of truth – that is of the concept of truth.”17
Here Hintikka ‘s reasons for rejecting the correspondence theory seem very
close to the reasons which led Wittgenstein to say that it is “impossible to describe the
fact which corresponds…to a sentence, without simply repeating the sentence.”18
But not only does Hintikka reject the correspondence conception in such a
passage, but he also seems to commit himself to what I have called the priority thesis.
The priority thesis, remember, is not a thesis about truth, but a thesis about meaning.
It says that there is no account of meaning independently from an antecedent grasp of
the meanings that we give to the sentences and expression of our language. This is
partly what is involved in Wittgenstein’s emphasis that meaning is, in a certain sense,
nothing but use. The only way in which we can spell out the meaning of our sentences
is by displaying how they are used in certain rule governed practices. Now, given that
among meaning-relations the relation between a sentence and its truth-conditions is
central, what would be the consequence of this for the truth-relation (or the truthproperty)? To say that a certain sentence is true is just to be able to use it within a
certain language game, the language game of assertion. But to know what truth “is” is
just to understand how the rule or rules for assertion function. And this kind of
knowledge seem only to be available from within our assertoric practices. In this
respect, Hintikka game-semantical ideas seem to be very close to Wittgenstein’s
version of the ineffability and priority theses. When he comments upon his gametheoretical account of truth conditions for first order sentences – in terms of Skolem
functions as above with (1) and (2) – Hintikka seems ready to draw the same
17
Hintikka 1996, p.44-45
Wittgenstein, Culture and Value, Oxford, Blackwell, 1980, p. 10 Quoted by Hintikka in “Is truth ineffable?”,
Selecte papers, 2, p.24.
18
10
conclusions about the use-character of the concept of truth as those that he draws
about the use-character of the concept of meaning:
“It may nevertheless be questioned whether the concept of truth in general is really
illuminated by the game-theoretical conditions. The job that they do is to specify what
quantificational sentences mean by specifying their truth conditions. the notion of truth is
here a mere auxiliary one, it seems. In other words, the first-ordrer semantical games
seem to be language games for quantifiers, and not for the concept of truth. This is
apprently in keeping with the nature of these games as games of seeking and finding.
The conceptual connection between quantifiers and the activities of seeking and
fining is easy to appreciate, but there does not seem to be any equally natural link
between semantical games and the notion of truth in general. this can be thought of a
being illustrated also by the impossibility of defining truth for quantificational
sentences in those first-order languages which receive their meaning from my
semantical games. On can suspect here, as Wittgenstein would have done, that the
concept of truth can only receive a use – and ergo a meaning – in the context of
certain other language games.” (Hintikka 1996, p.31-32)
The very fact, however, that we cannot spell out the meaning of the concept of
truth from without its use and from without our language games does not imply that
the concept of truth is ineffable or inexpressible. Actually when he comments upon
his truth-definitions for IF languages, which can be given from within these languages
and not from without in a distinct metalanguage, Hintikka emphasises that this feature
dispells the “myth that the notion of truth for a sufficiently strong language is
inexpressible in that language itself” and hence the ineffability thesis about truth. But
what is interesting is that Hintikka recruits this point in favour of the priority thesis.
He raises precisely the Wittgensteinian point about semantical games:
“How can these very same games also serve to give an altogether different kind
of concept its meaning, namely the concept of truth – at least the notion of truth as
applied to first order languages. How can one and the same language game serve to
lend a meaning to two different kinds of concepts, one of which (the concept of
truth) seems to be a metalogical one? This two hats problem can also be called
Wittgenstenstein’s problem (cf. here Hintikka 1986, ch.1). For Wittgenstein insisted
that you cannot speak meaningfully and nontrivially of the truth of the sentences of a
language in that language itself. Or since for Wittgenstein there is ultimately one
language (“the only language that I understand”), we cannot speak of truth
nontrivially, period. What looks like a metalogical discourse pertaining to the truth
and falsity of a fragment of language is for Wittgenstein a different “calculus”, a
11
different language based on a different language game. How then can the meaning of
first-ordrer languages be constituted by the csame language games? Doesn’t speaking
of truth take us ipso facto to a metatheoretical level ?” (Hintikka 1996, p.127-128)
And his answer to these questions is that giving the meaning for various
expressions (here quantifiers) is not the same a giving definitory rules; it is giving strategic
rules for the sentences which contain these expressions, and what it is to understand
the meanings of these expressions is just what it is to understand the the concept of
truth for the language which contain them. Thus the step from an understanding of
the expressions to an understanding of the truth of the sentences which contain them
is not a step “to a metalogical level”: “understanding the strategies available to the
players of a semantical game…[is] just what is needed to understand the concept of
truth” (ibidem, p.128). In other terms, the definability of truth – which refutes the
ineffability thesis – does not imply that our understanding of the concept of truth is not
implicit to our mastery of our language, hence it is compatible with the the priority
thesis.
If this is correct, Hintikka’s rejection of the ineffability thesis about truth and
meaning is still compatible with a form of minimalism. Contrary to the universalist
tradition in logic and semantics, Hintikka takes truth to be definable. Contrary to the
deflationist and disquotational conceptions of truth, he does not take the meaning of
the truth-predicate is to be exhausted by Tarski’s T-schema. But he holds in order to
understand the concept of truth, we do not have to use ressources which would
exceed our own grasp of the rules of our language. As I have suggested, although
Hintikka rejects the ineffability thesis, he still subscribes to a form of the priority
thesis. Now the priority thesis is does not, by itself entail any minimalism about truth,
but it does entail a mininalism about meaning, in the following sense: in grasping a
language rule, I grasp its applications, but I do not grasp any further rules determining
what its applications to particular cases consists in. Similarly when I understand a
strategic rule for a semantical game, I do not need further rules to grasp their
applications. Minimalism about meaning does not, by itself justify minimalism about
truth. But there is a straighforward incompatibility between a ( non minimalist)
12
conception of meaning in terms of truth-conditions and the minimalist theory of
truth, which has been spelled out by Dummett a long time ago:
“In order that someone should gain from the explanation that P is true in such
and such circumstances an understanding of the sense of P, he must already know
what it means to say that P is true. If he enquires into this he is told that the only
explanation is that to say that P is true is the same as to assert P, it will follow that in
order to understand what is meant by saying that P is true, he must already know the
sense of asserting that P, which was precisely what was supposed to be explained to
him.”19
Dummett actually uses the same point against a truth conditional conception of
meaning when he claims that such a conception can only lead to a minimalist (or
“modest”) conception of meaning. On a truth-conditional conception of meaning –
in particular when it takes the form of the Tarski-like requirement that truth
conditions be given by such T-sentences as :
(3) “Theetetus flies” is true (in English) if and only if Theetetus flies
what it is to know what “Theeetetus flies” means consists in knowing that (1) (on the
basis of its structure) expresses a truth. But of course one can know that this
metalinguistic sentence expresses a truth without knowing what the object-language
sentence “Theetetus flies” means, or the proposition that it expresses. In order to
know the meaning of “Theetetus flies” through (1), I must already know what
meaning of this sentence (on the right-hand side). This why Dummett says that the
attempt to specify what a speaker understands through (T) sentences like (1) can only
yield a modest theory of meaning, one which “is not intended to convey the concepts
expressible in the object-language, but to convey an understanding of that language to
one who already possesses those concepts”.On the contrary, a “rich” or “fullblooded” theory should “in the course of specifying what is required for a speaker to
19
Dummett , “Truth” (1959) in Dummett 1978, p.7.
13
grasp the meaning of a given word…explain what it is to possess the concept it
expresses”.20
Of course Hintikka is not a “modest” theorist of meaning in the sense in which
Dummett considers that Davidson is one; neither is he a minimalist theorist of truth.
But he is no more a full-blooded theorist in Dummett’s sense (nor, as we have seen,
would he suscribe to a constructivist conception of truth). Nevertheless, if I am right,
he shares Wittgenstein’s view that grasping the meaning of an expression is not
grasping a language-independent concept that this expression expresses. There is no
more to grasp of meaning than grasp of the strategic rules which are immanent to our
implicit understanding of our own language. In this respect, in accepting a version of
the priority thesis, Hintikka has not completely withdrawn the thesis of the
universality of language.
5.
I am not sure that he has completely withdrawn a (certain form of) minimalist
conception of truth either. In order to see this, let us first contrast a genuine
minimalist conception of truth, such as Horwich’s deflationism, with a more
substantive conception. According to Horwich (1990), the meaning of the truth
predicate, and the nature of truth itself, are completely exhausted by (a form of the)
the disquotational schema
The proposition that p is true iff p
The problem with this view is that there is more to truth than that. Truth does not
simply register the fact that we make certain assertions, the contents of which are
either quotable with the predicate “is true” or disquotable when one drops this
predicate. Truth registers a distinctive norm, which has a definite content (Wright
1992, Engel 2001, 2002). The norm in question is that our assertions, when true, are
not simply subject to disquotation and to the discipline of syntax, but also to certain
standards of objective correctness. The fact that a speaker who makes an assertion is
20
Dummett “What is a theory of meaning (I) 1974, in S. Guttenplan, ed. Mind and Language, Oxford, Oxford
University Press, p. (1993, op cit p. viii and p. 22. Sq.))
14
supposed to be justified in making it, and liable to answer quaries about it, is an
important fact which shows that truth has a more substantive content than what the
deflationist conception allows. The recognition of this fact does not imply that truth is
a substantive concept in the sense of the realist or anti-realist conceptions of truth,
such as correspondence, verificationist, or cohrence conceptions. But it implies that
our use of the truth predicate carries an implication that our statements can be
objective, answerable, and that speakers can potentially converge on them. In other
terms, the concept of truth is substantial in the sense that in use it, we commit
ourselves to a minimal form of realism. But this realism need not be of a metaphysical
kind, as in a correspondence theory of truth couched in terms of facts or states of
affairs. We only need to recognise that when someone asserts a certain sentence to be
true, he carries the implication that its content is knowable. The norm of assertion is
not so much truth than knowledge: in making an assertion I make a claim to knowledge,
and I do not simply express my belief that the assertion is true. Given that knowledge
implies truth, it is open to us to say that truth is the norm of assertion, through its
aiming at knowledge.21 I cannot here to develop these ideas (see Williamson 2000,
Engel 2002). But if they are correct, there is room for a form of minimalism about
truth – which would grant that there is not much more to truth than the
disquotational feature – which would nevertheless be substantive in the sense that it
registers the norm of knowledge.
At first sight, and on the one hand, it seems to me that Professor Hintikka
should agree with the view that I have just sketched. The idea that truth is not just
disquotation and that it registers a distinctive norm is but a version of Dummett’s
famous remark that truth is what our assertions aim at, just like winning is what our
playing a game aim at. Hintikka (1996, p.27) agrees with this analogy in so far as it is
understood as the analogy between truth and the existence of a winning strategy. And
we have seen that this reading is compatible with a form of realism, although not of
the metaphysical kind. On the other hand, I suspect that he would disagree, and
would say that my reading into the concept of truth a commitment to a knowledge
21
These issues have been discussed in Hintikka’s classic Knowledge and Belief (1962). But as far as I know, he
does not defend the view that assertion implies a claim to knowledge.
15
claim belies a confusion – already alluded to above- between semantical games, which
give the truth conditions of the relevant sentences of the language, and interrogative
games, which give the conditions of our reaching knowledge through inquiry.22 The
former are prior to the latter, and more fundamental. They characterise truth in
general, and not the way we come to know truth. Hintikka denounces the confusion
between semantic and interrogative games in the context of a criticism of the
verificationist conception of truth, which in some sense equates truth of knowledge of
truth. This is not, however, the point that I put forward when I say that truth registers
a norm of knowledge. It is not meant to say that truth is in some sense epistemic,
since the conception of truth which is here presupposed is realistic. But the
distinction that Hintikka makes between semantic games, dealing with truth, and
interrogative games, dealing with knowledge, shows that he is closer to a minimalist
conception of truth than the conception that I have advanced. Another sign of this
would be his recent claim (2002a) that in epistemology we need neither the notion of
knowledge nor the notion of belief, but only the notion of information. In so far as
Hintikka wants to dissociate the notion of truth from the notion of knowledge, and
epistemology from this very notion, it seems to me that he is prepared to adopt a
relatively thin concept of truth.
Let us, finally turn to the credentials of what I have called the priority thesis. The
priority thesis seems compelling only if the alternatives to an account of language
understanding from within our language games are a Platonist conception of concepts
or a psychologistic account of them. These were actually the two kinds of view that
Wittgenstein meant to attack in the rule following considérations. But these are not
the only options. If we could give a better account of what it is to possess a concept
than a Platonist or a psychologistic one, the priority thesis will seem less attractive.
Such an account would have to be substantive, in the sense that it would not explain
our grasp of a meaning or of a concept in terms capacities which would presuppose a
priori grasp of these meanings. On a minimalist theory of meaning, we do not in any
sense explain the meaning of “Theetetus flies” through a sentence like (1), because
we must already know what the right-hand side means. Similarly when a minimalist
22
For this distinction and the confusion in question, see for instance Hintikka 1996, p.36-37
16
conception of meaning is formulated in terms of language rules. In order for a
conception of meaning to be substantive, one has to frame in terms of necessary and
suffient conditions which to not presuppose the meaning of the target sentences,
along the lines of something like ( for instance, for the rule for assertion) :
(A) S is correctly used to make an assertion in language L if and only if
…
Now a game-theoretical semantics is just what can provide us with what we need on
the right hand side. But if what precedes is correct, even though game-theoretical
semantics uses such notions as that of a verifier or of a winning strategy, these are
notions which a speaker is supposed to already understand, and hence which are
redescriptions of what a user of language already knows. A better account seems to be
directly in terms of what a speaker knows:
(A)One correctly uses a S to make an assertion if and only if
one in justified in believing the proposition expressed by that use, that
it is true.
And here again the account yields an analysis only if the relevant notion of “justified
belief”, or indeed of knowledge spelled out. On such an account, meaning is
understood in terms of epistemic capacities or in terms of epistemic norms. To grasp the
meaning of an expression is to grasp a certain concept individuated in terms of its
cognitive role.23 Here too I cannot develop this proposal, but I suspect here again that
Professor Hintikka would say that it is to confuse semantical games and interrogative
games, truth conditions and knowledge conditions. But this too would show the
extent of his commitment to the priority thesis.
Hintikka’s discussions on truth are shaped by his concern to give an overall
account of the languages of logic and mathematics on the one hand, and of the
semantic structure of natural laguages on the other. I have not discussed his particular
proposal for a truth definition, but only his remarks on our “normal” concept of
23
For such views, see in particular Peacocke 1992, Skorupski 1997.
17
truth. Let us grant that truth is indeed definable, along the lines that he has proposed.
The ineffability thesis would then be disproved. But I have suggested that Hintikka is
still committed to the view that “one cannot use language to get outside language”,
with respect both to truth and meaning.
University of Paris-Sorbonne
september 2002
18
REFERENCES
Armstrong , D. 1997 A World of States of Affairs, Cambridge, Cambridge
University Press
David, M. Correspondence and Disquotation, Oxford, Oxford University Press 1994.
Dummett, M. 1978 Truth and other enigmas, Duckworth, London
Dummett, M. 1974 “What is a theory of meaning (I) 1974, in S. Guttenplan,
ed. Mind
and Language, Oxford, Oxford University Press,
Engel, P. 1988 “Plenitude and Contingency : Modal concepts in Nineteenth
Century
French Philosophy”, in S. Knuuttila ed. Modern modalities, Kluwer,
Dordrecht, 1988, pp.179-237
2001 “ Is truth a Norm ? ” in P. Pagin, G. Segal & P. Kotatko, eds
Interpreting
Davidson, Stanford, CSLI., 2001, pp. 37-51
2002 Truth , Bocks: Acumen
Hintikka, J. 1962 Knowledge and Belief , Ithaca: Cornell University Press
1976 "The Prospects of Convention T". Dialectica, 30, pp. 61-66.
1981 “Semantics: A Revolt against Frege”, in Floistad, G. (ed.)
Contemporary Philosophy, vol. 1, The Hague, Nijhoff, p.57-82
1987 “Self Profile”, in R. Bogdan , ed. Jaakko Hintikka, Profiles ,
Dordrecht
Reidel, pp.33-39
1989 “Is Truth Ineffable ?” in Les formes actuelles du vrai, entretiens de
Palermo, ed. N. Scardona, Enchiridion, Palermo, 1989, repr. in
Hintikka
1997, p.20-45 (quoted after this edition)
1991 « Defining Truth, the Whole Truth, and Nothing but the
Truth »,
Reports from the Departement of philosophy, Helsinki, 1, repr.
in 1997,
pp.46-97 (qoted after this edition)
1996 The Principles of Mathematics Revisited, Cambridge, Cambridge
University Press
1997 Lingua Universalis vs Calculus Ratiocinator: An Ultimate
presupposition in Twentithe Century Philosophy, Selected Papers, vol. 2,
Kluwer, Dordrecht
1997a “Contemporary Philosophy and the Problem of Truth”, in
1997, p.1-19
1998, Language, Truth and Logic in Mathematics, Selected Papers, vol 3,
Kluwer, Dordrecht
2001 “Post-Tarskian Truth”, Synthese, 2001, 126, 17-36
2002 ms “Truth, Negation and other basic notions of logic”
19
2002 a ms “Epistemology without knowledge and without belief”
Hintikka, J. and Hintikka, M. 1986 Investigating Wittgenstein, Blackwell, Oxford
Horwich , P. 1990 Truth, Blackwell, Oxford, 2nd ed. 1998
Peacocke C. 1992 A Study of Concepts, Cambridge Mass, MIT Press
Skorupski, J. 1997 “Meaning, use, verification”, in Wright, C. and Hale, B, eds, A
Companion to the Philosophy of Lanaguage, Oxford: Blackwell, pp. 29-59
Williamson, T. 2001 Knowledge and its limits, Oxford: Oxford University Press
Wittgensteinn, L. 1975, Philosophical Remarks, Oxford: Blackwell
Wright, C. Truth and Objectivity, Oxford, Oxford University Press, 1992